Μελέτη των εμφανίσεων μεγάλων αποζημιώσεων μέσω της θεωρίας ακραίων τιμών, με εφαρμογές στις αντασφαλίσεις
A study of appearances of large claims via extreme value theory, with applications in reinsurance
View/ Open
Keywords
Reinsurance ; Block Maxima ; Extreme value theory ; Peaks over Threshold (POT) ; Generalized Extreme Value Distribution (GEV) ; Generalized Pareto Distribution (GPD)Abstract
As extreme, can be considered the events, which are rare and intense (i.e. appear with very small probability and their occurrence may have high impact in the model we study). According to Extreme Value Theory these extreme events should be studied separately from the entire dataset in order to extract useful results about their behavior. These results will help us find ways to deal with (i.e. to normalize) the impending consequences in case an extreme event occurs. One of numerous fields in which Extreme Value Theory (EVT) can be applied, is insurance risk models, where e.g. we may consider an insurance portfolio which could be hit by the occurrence of very large (i.e. extreme) claims. In actuarial practice, reinsurance is used by insurance companies as a measure of the risk management process. The purpose of this thesis is to connect the Extreme Value Theory and the concept of Reinsurance. By studying and interpreting appropriately the results of extreme losses through Extreme Value Theory, an insurance company could arrange a reinsurance contract by estimating the corresponding premium. Initially we present several types of Reinsurance contracts and then we offer an introduction to the “Block Maxima” and “Peaks Over Threshold” (POT) methods in a simple and comprehensible way through a numeric example. In the final chapter we exploit specific results of extreme value theory in order to study the problem of pricing a non-proportional reinsurance contract.